A Verified SAT Solver Framework with Learn, Forget, Restart, and Incrementality (Extended Abstract)

Jasmin Christian Blanchette (1, 2, 3) , Mathias Fleury (1, 2) , Christoph Weidenbach (1, 2)

1 MPII - Max-Planck-Institut für Informatik
2 VERIDIS - Modeling and Verification of Distributed Algorithms and Systems
3 MOSEL - Proof-oriented development of computer-based systems

We developed a formal framework for CDCL (conflict-driven clause learning) in Isabelle/HOL. Through a chain of refinements, an abstract CDCL calculus is connected to a SAT solver expressed in a functional programming language, with total correctness guarantees. The framework offers a convenient way to prove metatheorems and experiment with variants. Compared with earlier SAT solver verifications, the main novelties are the inclusion of rules for forget, restart, and incremental solving and the application of refinement.

Domaines: Logique en informatique [cs.LO]

Jasmin Christian Blanchette, Mathias Fleury, Christoph Weidenbach. A Verified SAT Solver Framework with Learn, Forget, Restart, and Incrementality (Extended Abstract). Isabelle Workshop 2016, Aug 2016, Nancy, France. ⟨hal-01401807⟩